Jump to content

PDP-11/34 Troubleshooting Guide

From RetroTechCollection
Modules in a PDP-11/34's Unibus backplane (Autopilot, CC BY-SA 3.0)

DEC's PDP-11/34 System User's Manual gives a seven-step quick verification routine for the PDP-11/34, with a flowchart for each step that names the likely problem areas; it works with either the KY11-LA operator's console and a terminal, or the KY11-LB programmer's console.[1] Run it after checking the power supply and the module configuration. Restorers have added two faults DEC did not list: failing solder joints on the backplane and the H765 power line monitor.[2][3]

Before you start

[edit | edit source]

Check DEC's installation rules for the 11/34 first:[4]

  • The KD11-E or KD11-EA modules go in the first two backplane slots.
  • The M9302 terminator goes in the last slot, and never in a modified Unibus slot (slots 2 to 8, sections A and B, on the DD11-PK and DD11-DK; slots 2 and 3 on the DD11-CK).
  • Every unoccupied SPC slot needs a G727 grant card in connector D, or grant continuity is lost.
  • A BC11-A Unibus cable must never be plugged into a modified Unibus slot.
  • In boxes other than the BA11-L the LTC L signal cannot drive several loads: enable the line clock on only one DL11-W and remove R63 from the others, or the line clock fails.

DEC adds two warnings: the DC OFF position of the power switch does not remove AC power from the system, and the DC ON lamp shows only that DC power is applied to the logic, not that it is within limits.[4]

With the H765 in the BA11-K, only the DC ON position of the switch works; STBY acts as off.[5][6]

Power supply checks

[edit | edit source]

Measure +5 V and −15 V at the backplane with an oscilloscope, ground lead on a backplane ground pin. DEC's limits for the regulators are ±5%, with ripple up to 200 mV peak to peak on +5 V (H744) and 450 mV on −15 V (H745).[7]

  • A regulator output near 0 V may be a crowbarred regulator. Power down fully, turn its adjustment half a turn anticlockwise, power up and readjust on the scope. If it crowbars again, look for a bad harness connector, a loose wire across backplane pins or a bent pin.[8]
  • No +15 V, no line clock, AC LO stuck: suspect the 5411086 power line monitor, whose unfused bridge rectifier can short and burn its edge-connector AC contacts.[3][9]
  • AC contactor buzzes and does not close at power-up: the 50 µF capacitor on the AC power control board has dried out.[3]
  • A regulator whines: suspect its output capacitors or its switching transistor.[10][11]

See PDP-11/34 Power Supply Restoration.

Quick verification routine

[edit | edit source]

The routine assumes the M9301 switches are all on. Perform the actions in order; if one fails, use the table for that action.[1]

Quick verification routine (EK-11034-UG-001, section 6.2.1)
Action Correct result
1. Turn power on Register printout on the console terminal; RUN lamp on
2. Load address 173024 and examine it Contents 173000
3. Type a boot command for a device that does not exist Register printout; RUN on (with an M9301-YF the processor halts instead)
4. Load address 0, examine, deposit 777, examine, start Keyboard stops responding; RUN on
5. Halt the processor RUN off
6. Continue RUN on
7. Halt, BOOT, return HALT/CONT to CONT (KY11-LA) Register printout with "OLD PC" 0

[1] At power-up the M9301 runs five processor tests (single-operand, double-operand, jump, and two non-modifying byte and operand tests) and then prints R0, R4, R6 and R5, R5 holding the old PC, followed by a $ prompt. A prompt shows that part of the processor, the M9301, the console interface and terminal, and the Unibus all work.[1]

Action 1: no printout at power-up

[edit | edit source]
Action 1 failures (EK-11034-UG-001, Figure 6-1)
Symptom Further test Possible problem
RUN flashes once, does not stay on Press and release BOOT/INIT; if the printout now appears, M9301 switch 2 is off HALT/CONT at HALT; console terminal interface missing or at the wrong address; HALT GRANT or HALT REQUEST hanging the bus (check BUS SACK on the processor); one of the first five tests failed; faulty M9301
RUN off, DC ON lit Generate BUS INIT (BOOT/INIT with CONT on the KY11-LA, or INIT with CNTRL on the KY11-LB); RUN should light if +5 V is present Console fault if RUN still stays off
RUN on, no printout, does not halt Power down, remove the M9302, power up Bus grant problem if the printout then appears: a missing grant card, or one fitted with its etch away from the processor. Otherwise check BUS SACK and the MSYN and SSYN lines
RUN on, halts and continues normally Power down, remove the console interface and fit a grant card, power up Console terminal or interface if RUN then goes off; otherwise a program loop or a failed M9301 test (branch to self)

[12] The NPG signal is carried between slots by backplane jumpers, not by grant cards.[12] Henk Gooijen describes the common case: the CA1 to CB1 wire in connector C of a slot carries NPG and must be cut when a DMA controller is fitted there; if that controller is later removed, the wire has to be restored or a G7273 grant card fitted. With the NPG chain broken the RUN lamp stays on and the processor will not halt from the console.[13]

Actions 2 to 7

[edit | edit source]
  • Action 2. If 173024 does not read 173000, the M9301 switches are set wrongly; if they are right, the M9301 has failed. Other odd responses are an input error (such as lower-case typing) or a console emulator failure.[14]
  • Action 3. If the RUN lamp goes off after booting a non-existent device, a diagnostic failed (or, on an M9301-YF, all passed). Halt, BOOT, return to CONT and read the OLD PC from the printout:
M9301 diagnostic failure addresses (EK-11034-UG-001, p. 6-12)
Result M9301-YA M9301-YB M9301-YF
Failed test 6 165320 173452 165650
Failed test 7 165350 173472 165664
Failed test 7 (second entry as printed) 165372 173514 165700
Failed memory test 165536 173764 166000
All tests successful – – 000002

[14] A failure at 165250 (YA), 173402 (YB) or 165600 (YF) means address 500 did not return SLAVE SYNC. After a memory test failure the printout holds the failing address and the expected and received data: R0 expected, R4 received, R6 address on the YA and YB; R0 address, R4 received, R6 expected on the YF. If expected and received data are the same, DEC suggests an intermittent timing or margin fault.[14]

  • Action 4. Keyboard still responds or RUN off: input format error or M9301 firmware fault.[15]
  • Action 5. RUN stays on after HALT: HALT REQUEST or HALT GRANT problem, or a faulty KY11-LA or KY11-LB.[15]
  • Action 6. RUN stays off after CONT: faulty console.[15]
  • Action 7. No printout: faulty or wrongly connected boot cable between console and M9301. Wrong OLD PC: noise on the boot cable, or BOOT/INIT pressed twice.[15]

Backplane faults

[edit | edit source]

Jörg Hoppe found an 11/34A whose NPG chain on the DD11-PK was open: the resistance between NPG pins wandered between 0 and 10 kΩ. The wire-wrap was intact; the fault was a wire-wrap pin that the solder had never wetted, so it made contact with the board track only when undisturbed. He estimated that 20–40% of the pins on that backplane were unwetted on at least one side, and found the machine passing console tests at first and then failing more often as boards were moved. Find such a pin by watching the resistance while moving each pin slightly.[2] Reflowing such pins, or replacing the backplane, are the cures he discusses.[16]

Memory faults

[edit | edit source]

MOS memory loses its contents when power is removed unless the H775 battery backup is fitted; core memory keeps them.[17] MOS boards take their refresh power from the +5B, +15B and −15B backplane buses. In a 10½-inch box without battery backup those buses must be jumpered to the normal rails, or they float and the MOS memory fails.[6] Each MOS board adds 0.5 A of refresh load to the +5 V regulator that feeds the +5B bus.[6]

Diagnostics

[edit | edit source]

DEC lists these diagnostics for the 11/34: MAINDEC-11-DFKAA (CPU), DFKAB (traps), DFKTG (memory management exerciser), DFKAC (EIS instructions), DZKMA and DZQMC (MOS and core memory exercisers), DCMFA (combined parity memory test), DZKAQ (power fail, where core or battery-backed MOS memory is fitted), DZDLD (DL11-W), and, where no DL11-W diagnostic is available, DZKLA (KL11/DL11-A) and DZKWA (line clock).[18][19] With the KY11-LB and its maintenance cables to the M7266, maintenance mode (CNTRL with key 1) single-steps the microprogram and displays the micro-PC; key 5 lets the console examine and deposit on the Unibus with no processor present.[20]

[edit | edit source]

References

[edit | edit source]
  1. ↑ 1.0 1.1 1.2 1.3 DEC, EK-11034-UG-001, section 6.2.1, pp. 6-2 to 6-5.
  2. ↑ 2.0 2.1 Jörg Hoppe, "Trying to fix a DEC PDP-11/34 backplane – diagnosis", retrocmp.com, retrocmp.com.
  3. ↑ 3.0 3.1 3.2 Jonathan Chapman, "Repairing the PDP-11/10 Power Supply", Glitch Works, 5 June 2019, users.glitchwrks.com.
  4. ↑ 4.0 4.1 DEC, EK-11034-UG-001, section 6.1, pp. 6-1 to 6-2.
  5. ↑ DEC, EK-11034-UG-001, section 3.2.3, p. 3-9. DC power switch with the H765.
  6. ↑ 6.0 6.1 6.2 Digital Equipment Corporation, PDP11/34A Power System, EK-1134A-TM-002, sections 2.5 to 2.7 and Table 2-1 (File:PDP-11-34A Power System Description (EK-1134A-TM-002).pdf).
  7. ↑ Digital Equipment Corporation, PDP-11 Mainframe Troubleshooting Guide, December 1976, DECAID PWRB-1 (File:PDP-11 Mainframe Troubleshooting Guide (December 1976).pdf).
  8. ↑ DEC, PDP-11 Mainframe Troubleshooting Guide, DECAID PWRB-3, crowbar problems.
  9. ↑ "DEC Power Line Monitor", Computer History Wiki, gunkies.org/wiki/DEC_Power_Line_Monitor.
  10. ↑ Rob Jarratt, "RE: Source for replacement caps in H744 regulators", cctalk mailing list, 6 January 2022, mail-archive.com.
  11. ↑ Jon Elson, "Re: Source for replacement caps in H744 regulators", cctalk mailing list, 6 January 2022, mail-archive.com.
  12. ↑ 12.0 12.1 DEC, EK-11034-UG-001, section 6.2.2 and Figure 6-1, pp. 6-6 to 6-9.
  13. ↑ Henk Gooijen, "PDP-11/34", pdp-11.nl. NPG wire, G727A and G7273 grant cards.
  14. ↑ 14.0 14.1 14.2 DEC, EK-11034-UG-001, Figures 6-2 and 6-3, pp. 6-11 to 6-12.
  15. ↑ 15.0 15.1 15.2 15.3 DEC, EK-11034-UG-001, Figures 6-4 to 6-7, pp. 6-13 to 6-14.
  16. ↑ Jörg Hoppe, "Trying to fix a DEC PDP-11/34 backplane – therapies", retrocmp.com, retrocmp.com.
  17. ↑ Digital Equipment Corporation, PDP-11/34 System User's Manual, EK-11034-UG-001, July 1977, section 1.2.6, pp. 1-7 to 1-8 (File:PDP-11-34 System User's Manual (EK-11034-UG-001).pdf).
  18. ↑ DEC, EK-11034-UG-001, section 5.8 and Table 5-1, pp. 5-23 to 5-24.
  19. ↑ DEC, PDP-11 Mainframe Troubleshooting Guide, DECAID MAINDA, which lists DZKAQ as the power fail test.
  20. ↑ DEC, EK-11034-UG-001, appendix A, KY11-LB maintenance mode, pp. A-1 to A-3.
Retrieved from "https://wiki.retrotechcollection.com/index.php?title=PDP-11/34_Troubleshooting_Guide&oldid=21545"